ILP Formulations for $p$-Cycle Design Without Candidate Cycle Enumeration

Abstract

The concept of <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">p</i> -cycle (preconfigured protection cycle) allows fast and efficient span protection in wavelength division multiplexing (WDM) mesh networks. To design <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">p</i> -cycles for a given network, conventional algorithms need to enumerate cycles in the network to form a candidate set, and then use an integer linear program (ILP) to find a set of <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">p</i> -cycles from the candidate set. Because the size of the candidate set increases exponentially with the network size, candidate cycle enumeration introduces a huge number of ILP variables and slows down the optimization process. In this paper, we focus on <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">p</i> -cycle design without candidate cycle enumeration. Three ILPs for solving the problem of spare capacity placement (SCP) are first formulated. They are based on recursion, flow conservation, and cycle exclusion, respectively. We show that the number of ILP variables/constraints in our cycle exclusion approach only increases linearly with the network size. Then, based on cycle exclusion, we formulate an ILP for solving the joint capacity placement (JCP) problem. Numerical results show that our ILPs are very efficient in generating <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">p</i> -cycle solutions.